Critical Power Exchange LLC - Liebert Air Conditioners, Power Distribution Units, UPS, Raised Access Computer Flooring, Diesel Generators, Halon Fire Suppression

Surplus power equipment from Critical Power Exchange

Availability - Savings Solutions - Resource Recovery - Asset Management - Project Management

CALL TODAY!
Toll-Free:
877.315.4176

Home > Articles > Standby Power

Articles

Standby Power
How Much Should Your Enterprise Have?

People in general may have an innate fear of the dark, but for IT managers, this impulse can reach sheer frenzy in a matter of minutes. Very few enterprises are comfortable with the idea of sitting in the dark with all operations halted for any length of time. And the more a company’s operations hinge on 24x7 online access, such as with ecommerce, the more that company is likely to be intolerant of any downtime and will take costly measures to prevent it.

“You’re just trying to carry enough backup power of any type—battery, gen sets, whatever—to keep your server room or data center running long enough to perform an orderly shutdown,” says Darin Stahl, lead analyst at Info-Tech Research Group. “The duration of that runtime is a function of business needs, criticality, and cost, and cost is a derivative of impact to business. I see a lot of SMEs out there using a standalone or a rack battery solution to keep critical servers up for 30, even 45, minutes.”

How Much Standby?
One of the biggest reasons enterprises get in trouble with standby power is failure to honestly and accurately assess this criticality. Many times, the systems that admins think just can't possibly be allowed to go down really can, especially when viewed against the expenses required to prevent downtime. People understand that accidents happen, especially to smaller enterprises that provide superior service instead of having massive infrastructure redundancy. Provided that systems are in place such that the most important servers can be shut down within a few minutes so that no data is lost and/or service can be failed over to a remote location, then an additional standby power investment may not be warranted.

Of course, this still leaves the question of how much standby power an enterprise really does need. Most require at least some, and deciding on the number of standby minutes is only half of the question.

“Typically, clients need to assess the critical loads in their facility that they need covered,” says Steve Luedtke, director of marketing for Critical Power Exchange (www.criticalpower.com). “They may not need all of the lights, but the cooling and blade servers or whatever will need to stay on, and they’ll need to figure out the kVA and amps for that pull. If they don’t already have one onsite, they should get a local electrical engineer in there to walk through [and decide] how everything should go together. You tell him what you think your critical needs are, then he can assess and set up the system as to how it would go.”

Coming up with this list may not always be as easy as it seems. How much of the data center infrastructure (routers, switches, storage, and tape systems, among other things) needs to be included? How much backup lighting? Does the cooling need to stay up as long as the servers do? Info-Tech’s Stahl also suggests that an audit of data center power lines should be done before arriving at this standby load estimate. Although data centers may start out with all plugs cleanly marked and accounted for, time and entropy often have their way. Systems move. A cord for this one thing was a bit too short, and, in a hurry, well. . . . Things that were intended to be on backup power were moved off, and things that were never considered critical are suddenly part of the standby power budget. The results can contribute to significant over- or under-budgeting on standby power.

Wonky Budgets
According to Critical Power’s Luedtke, the average SME standby power solution he sells today is 1,000kVA. He also notes quoting a lot of backup products in the 1 to 1.5kVA range to clients who have under-budgeted their standby needs and can’t afford large-name OEM solutions. Why so many under-budgets? Luedtke says it’s “the age-old problem where they wrote the bids two years ago, and since then, the prices went up.” Especially in the current economy, in a business where lead times can be long and inventories short, some vendors may take the chance to squeeze for higher margins. Engineers need to take into account how far out the bid is and allow for extra expenses . . . and managers need to pad in a certain percent for tough times ... just in case.

On the other hand, many enterprises radically over-budget their standby power needs. Greg Fournier, director of IT power and distribution solutions, product management, at APC (www.apcc.com), says there are three basic models for buying standby power. First, small enterprises will often purchase single-phase distributed UPSes, meaning a new UPS for each new server that gets added.

“The second, more historical way is enterprises try to look 10 years in the future and buy a three-phase UPS for that time,” says Fournier. “They run it inefficiently for the first five years and hopefully grow into the capacity and efficiency they expected. You oversize your UPS based on uncertainty.”

The problem with this approach is that, just like power supplies, standby power systems perform less efficiently when operating outside of their peak utilization zones. Fournier notes that a 10-year UPS is likely going to run at 20 to 30% capacity on Day 1, "so at such low utilization, you're losing 20% of the energy to waste heat." Obviously, this magnifies the total power cost because the business not only has to pay for energy into the standby system but also all of that extra waste heat from excess inefficiency.

The third option is a modular approach. This entails buying a similar sort of “big steel” standby power framework used in 10-year solutions, but the battery systems are modular, so enterprises can purchase capacity as they need and grow. If the object is to stay near 80% utilization, an enterprise could start at 70%, grow up to 85% or 90%, then add another module to knock the capacity back down to 70+%.

Time To Re-Evaluate
There is no common rule for how often an enterprise needs to analyze and recalibrate its standby power usage. One advantage of the buy-a-server, buy-a-UPS model is that there’s little recalibration ever required. Provided that the server configurations don’t change and batteries are monitored regularly (at least once per month), standby power should remain well-fit. Keep in mind that UPS batteries need replacing every 24 to 36 months, on average.

For larger deployments in which standby systems are centralized, an annual site reanalysis is more appropriate. Based on this, forward-looking plans for the next six months to three years should be fine-tuned. Just remember to integrate plans for upcoming equipment changes—both additions as well as consolidations—that could affect total capacity numbers.

Key Points

Enterprises often under-budget total standby power costs and over-budget standby power capacity needs.
Older standby power models in which enterprises bought one solution meant to last for a decade can result in significant inefficiency and wasted resources. A pay-as-you-go model could make more sense.
Reanalysis of an organization’s standby power implementation should be done at least annually, with the findings being used to tweak capacity to optimal levels.

Critical Power Exchange, LLC
3808 N. Sullivan Rd, Building 29A
Spokane Valley, WA 99216

Toll Free (877) 315-4176
Phone (509) 228-0178
Fax (509) 228-0378
E-Mail info@criticalpower.com